Uncertainty-aware waveform modeling for high signal-to-noise ratio gravitational-wave inference

Semianalytical waveform models for black hole binaries require calibration against numerical relativity waveforms to accurately represent the late inspiral and merger, where analytical approximations fail. After the fitting coefficients contained in the model are optimized, they are typically held fixed when the model is used to infer astrophysical parameters from real gravitational-wave data. Though point estimates for the fitting parameters are adequate for most applications, they provide an incomplete description of the fit, as they do not account for either the quality of the fit or the intrinsic uncertainties in the numerical relativity data. Using the imrphenomd model, we illustrate how to propagate these uncertainties into the inference by sampling the fitting coefficients from a prior distribution and marginalizing over them. The prior distribution is constructed by ensuring that the model is compatible with a training set of numerical relativity surrogates, within a predefined mismatch threshold. This approach demonstrates a pathway to mitigate systematic bias in high signal-to-noise events, particularly when envisioned for the next generation of semianalytical models.